Monday, October 3, 2011

Reading: (Spatial Vis) A survey of algorithms for volume visualization

The article says that it will detail about 5 different algorithms that underpin volume visualization, but since it is from the 1991 SIGGRAPH, I wonder how many of them still apply today. It actually says that volume visualization "usually" applies to scientific visualization, which I suppose is probably true, since "science" is a very large field. The article mentions that animation is key to volume visualization, although I wonder if that just means interaction with rotations, pans and zooms.

The article mentions two major categories of volume visualization that all algorithms fall into: direct volume rendering (DVR) and surface-fitting (SF), and mentions several advantages and disadvantages for that family of algorithms. The article then mentions 11 algorithm implementations. To see how useful they all may still be, I did a quick round of googling, and apparently ray-casting, marching cubes, and splatting may be the most well-known, at least.

Finally, the article concludes by noting that volume visualizations are so important to get right because of the many critical applications where they are made to perform, such as medical diagnosis. If the visualization is inaccurate, it could lead to an incorrect diagnosis, so care must be taken, and a substantial amount of validation should take place.

All in all, this was a good paper to use to get an overview of volume visualization. I assume at least some of the algorithms that were discussed are still in use.